February 2022 Summaries

At AWS re:Invent 2021, LaunchDarkly engineers Mike Zorn and Justin Caballero presented techniques for painless migration using feature flags. They demonstrated how to convert a streaming event architecture and migrate production databases with zero downtime through gradual, reversible, and verifiable processes controlled by feature flags. The talk covered two examples of larger migrations undertaken at LaunchDarkly: the migration of a high-throughput data processing system and the migration of their core data store. Feature flags were used to change runtime behavior without any other modification, allowing for low latency updates and enabling additional use cases such as monitoring fidelity changes and handling migrations safely while mitigating risks.

The recent State of Feature Management report revealed that Slack is the most popular integration for LaunchDarkly customers, with nearly 40% relying on it. This popularity can be attributed to Slack's widespread use among distributed teams during the pandemic. LaunchDarkly's Slack app allows users to view, monitor, and control feature flags from their team's workspace. To get more value from these notifications, some best practices include reducing noise by adjusting filters and understanding channels, determining notifications based on specific events in LaunchDarkly, sharpening notification filters for greater specificity, changing channels to ensure the right audience receives relevant information, and leveraging links to provide context and centralize feature information.

TypeScript has seen a significant rise in popularity and is currently the seventh most popular language, with developers expressing interest in learning it. It is a superset of JavaScript that adds type checking capabilities. In this article, we discuss how to integrate LaunchDarkly's Node.js SDK into a TypeScript project and the React SDK for frontend use in a TypeScript project. By using types provided by the SDK, developers can incorporate type checking throughout their code. The integration process involves importing necessary modules, initializing clients, and handling flag values.

Paramount Tech Content Engineering team uses LaunchDarkly, a feature management platform, to improve their software delivery process and increase developer productivity. Before using LaunchDarkly, the team faced challenges such as long release cycles, coordinating multiple teams' releases, and slow bug fixing times. With LaunchDarkly, they can now deploy unfinished features without exposing them to production traffic, remotely disable flags tied to broken features at runtime, and control feature releases for specific users or environments. This has led to a 100x improvement in deployment frequency and faster resolution of production issues. Feature management is considered "one of the three legs of the stool" alongside CI/CD and automated testing in their software delivery process.

LaunchDarkly has introduced a new visual tool called Workflow Builder to automate feature flag rollouts. This feature is designed for teams with pre-defined release plans and aims to reduce the risk of accidental flag changes, allowing more time for planning, development, and innovation. The Workflow Builder enables users to create customized workflows with multiple approvals, user segment targeting, and flexible timelines. New features include out-of-the-box progressive rollout options, a read-only view for ongoing workflows, and granular targeting actions. The Workflow Builder is available on all Enterprise plans and can be accessed through the Workflows tab of any feature flag.

LaunchDarkly, a feature management platform, uses feature flags not only for feature toggling and A/B testing but also to manage various aspects of its product. The company recently undertook a major project to centralize and standardize the way it surfaces complex feature sets to different account types. By using a single boolean flag with four boolean layers, each managed by custom properties, LaunchDarkly was able to create a flexible system for managing integration offerings based on customer plans. This approach allows for easy extensibility, minimal documentation requirements, and no need for new PRs in the product code when adding new integrations.

In a recent chat with LaunchDarkly CEO, Edith Harbaugh, DevOps pioneer Patrick Debois discussed the potential pitfalls of over-reliance on specific methodologies like DevOps. He emphasized that while these practices can be beneficial, they should not be seen as cure-alls for all problems in software engineering. Debois also highlighted the cyclical nature of movements within the development world and how they often get reduced to a set of practices people remember them for. The conversation touched on the importance of incremental progress, staying open to learning, and embracing pliability in processes to achieve long-term, generative growth.

Serverless computing has revolutionized application development by eliminating the need for maintaining server farms or scaling applications manually. However, serverless applications often rely on numerous API calls, leading to latency and performance bottlenecks due to geographical distances between clients and servers. Edge computing aims to solve these issues by deploying critical portions of application code onto servers across multiple regions globally, reducing latency for many aspects of the application. This post explores popular edge function solutions like AWS Lambda@Edge and Cloudflare Workers, which are replicated across multiple regions globally. It also demonstrates how integrating these solutions with LaunchDarkly flags can enable powerful feature management and A/B testing capabilities without any client-side code. Lambda@Edge functions work similarly to regular Lambda functions but have some limitations on region, environment variables, and runtime. Users can integrate LaunchDarkly within a Lambda@Edge function using the server-side Node SDK. Cloudflare Workers are also serverless functions that work like AWS Lambda functions, with every Worker being an edge function deployed to Cloudflare's global network of CDNs. By integrating feature management and feature flags into edge functions, developers can quickly and transparently release features, change logic, adjust the HTML response, and more with just a switch in the LaunchDarkly dashboard. This enhances user experience by offering faster and better-performing applications while also enabling new possibilities for application functionality.